Proc. Natl. Acad. Sci. USAVol. 84, pp. 5340-5344, August 1987Genetics

Differential effects of base-pair mismatch on intrachromosomalversus extrachromosomal recombination in mouse cells

(homologous recombination/gene conversion/herpes simplex virus thymidine kinase gene/repeated sequences)

ALAN S. WALDMAN AND R. MICHAEL LISKAY

Departments of Therapeutic Radiology and Human Genetics, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06510

Communicated by Franklin W. Stahl, March 27, 1987

ABSTRACT To initially determine the effect that base-pair mismatch has on homologous recombination in mamma-lian cells, we have studied genetic recombination betweenthymidine kinase (tk) gene sequences from herpes simplex virus1 and 2. These tk genes are =81 % homologous at the nucleotidelevel. We observed that, in mouse LTK- cells, intrachromo-somal recombination between type 1 and type 2 tk sequences isreduced by a factor of at least 1000 relative to the rate ofintrachromosomal recombination between homologous type 1tk sequences. In sharp contrast, the rate of intermolecular orintramolecular extrachromosomal recombination between theheterologous tk sequences introduced by calcium phosphate ormicroinjection was reduced only by a factor of3 to 15 comparedwith extrachromosomal homologous tk crosses. Our resultssuggest differences between the mechanisms of extrachromo-somal and intrachromosomal recombination in mammaliancells.

To elucidate the mechanism of homologous recombination inmammalian cells, investigators have studied genetic recom-bination using both extrachromosomal (1-10) and intrachro-mosomal (1, 11-14) systems. One important issue to addressregarding homologous recombination is how the extent ofhomology between sequences influences the rate of recom-bination. Two groups have addressed this issue usingextrachromosomal mammalian recombination systems (15,16), whereas our laboratory has used an intrachromosomalsystem (17). The general finding is that when two sequencesshare several hundred base pairs (bp) of homology, the rateof recombination is proportional to the amount of homology.When the homology is reduced below -200 bp, the recom-bination rate drops off rapidly. In contrast to mammalianhomologous recombination, prokaryotic recombination ap-pears to require only -50 bp of homology to proceedefficiently (18-20).

In each of the above studies of the homology requirementsof mammalian recombination, the rate of recombination wasdetermined as a function of the extent of sequence overlap.Another way to study the homology requirements of recom-bination is to determine the effect that base-pair mismatchhas on the recombination rate. It has been shown that thehomologous recombination machinery of Escherichia coli isvery sensitive to base-pair mismatch, with 16% mismatchresulting in a decrease by a factor of 100 in the rate ofphage-plasmid recombination (19). In E. coli, the crucialfactor that determines the recombination rate appears to bethe length of uninterrupted stretches of homology (19, 20). Inmammalian cells, it has been shown that recombinationbetween the genomes of adenovirus types 2 and 5 can occurwithin regions exhibiting 90% sequence homology, but notwithin regions exhibiting 47% sequence homology (21).

As a further step toward understanding the homologyrequirements of recombination in mammalian cells, we havestudied genetic recombination between thymidine kinase (tk)gene sequences from herpes simplex virus 1 and 2 (HSV-1and HSV-2, respectively). These tk genes are 81% homolo-gous at the nucleotide level (22, 23). We have studiedrecombination between tk gene sequences using an intra-chromosomal system and two extrachromosomal systems.The data presented in this report show that intrachromosomalrecombination is reduced by more than a factor of 1000 by19% base-pair mismatch, whereas extrachromosomal recom-bination is reduced by a factor of only 3-15.

MATERIALS AND METHODSCell Culture and Transfections. Mouse L cells deficient in

thymidine kinase (LTK- cells) were cultured and transfectedby the calcium phosphate coprecipitate method (24),electroporation (25), or microinjection (26), as previouslydescribed.

Determination of Extrachromosomal Recombination Rates.Cells were transfected by the calcium phosphate coprecipi-tate method (24) or by direct microinjection (26) into thenucleus and were allowed to recover from the treatment for-20 hr in Dulbecco's modified Eagle's medium supplement-ed with 12% fetal calf serum (GIBCO). At this point, the cellswere refed with media containing hypoxanthine/aminopter-in/thymidine (HAT) (27). HAT-resistant colonies (HATr)were counted 14 days later, as a measure of the level ofrecombination to produce tk' cells that had occurred duringthe 20-hr period after transfection or microinjection.

Determination of Intrachromosomal Recombination Rates.Fluctuation analysis tests were done as described (28).

Plasmid Constructions. The vector used in all constructionsis a derivative of pSV2-neo (30) with restriction site alter-ations as described (28). Xho I linker insertion mutations ofthe HSV-1 (strain F) thymidine kinase (tk) gene were giftsfrom D. Zipser and J. Kwoh (Cold Spring Harbor Labora-tory). The HSV-2 (strain 333) tk gene was a gift from D.Galloway. Mutant no. 8 contains an 8-bp Xho I linker insertedat nucleotide 1215 of the HSV-1 tk gene, whereas mutant no.28 contains an 8-bp Xho I linker inserted at nucleotide 1036of the HSV-1 tk gene [numbering according to Wagner et al.(31)]. Plasmid pTK1-8 contains the no. 8 mutant tk geneinserted on a 2.4-kb BamHI fragment into the unique BamHIsite of the pSV2-neo vector after attachment of BamHIlinkers (New England Biolabs); plasmid pTK1-28 is a similarconstruct containing the no. 28 mutant tk gene on a 2.4-kbBamHI fragment.The 800-bp EcoRV-Stu I restriction fragment of the HSV-2

tk gene was isolated. This fragment is missing the 30% of the

Abbreviations: HSV-1, herpes simplex virus 1; HSV-2, herpessimplex virus 2; HAT medium, hypoxanthine/aminopterin/thymi-dine-containing medium that selects against tk- cells; HATr, HAT-resistant colonies (tk+).

5340

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 84 (1987) 5341

coding region of the HSV-2 tk gene that maps upstream fromthe EcoRV site and does not contain the polyadenylylationsignals of the HSV-2 tk gene that map downstream from theStu I site (22, 23). Using HindIII linkers (New EnglandBiolabs), this fragment was then inserted into the uniqueHindIII sites of the pSV2-neo vector, pTK1-8, and pTK1-28to produce, respectively, plasmids pTK2, pTK2TK1-8, andpTK2TK1-28.Plasmid pAL5 is identical to pTK2TK1-8 except that it

contains the 1.2-kb HincII-Sma I fragment of the HSV-1(strain F) tk gene inserted at the HindIII site, using HindIlllinkers. The fragment encodes most of the coding region butlacks the tkgene promoter and polyadenylylation signals (31).Plasmid pAL2 contains the 1.2-kb HincII-Sma I fragment ofthe HSV-1 tk gene inserted into the HindIII site of thepSV2-neo vector.Southern Hybridization Analysis. DNA isolation and

Southern blotting analysis were accomplished as described(28).

RESULTS

Intrachromosomal Recombination Between HSV-1 andHSV-2 tk Sequences. We have studied intrachromosomalrecombination between tk gene sequences from HSV-1 andHSV-2. These genes are 81% homologous at the nucleotidelevel, whereas the proteins are 75% homologous (22, 23). Thenucleotide mismatches between the genes are fairly evenlydistributed, with the longest stretches of perfect sequencematch being <30 bp. Two plasmids, pTK2TK1-8 andpTK2TK1-28, containing two different Xho I linker insertionmutations of the HSV-1 (strain F) tk gene were constructed.These plasmids also contain a defective fragment of thewild-type tk gene from HSV-2 (strain 333) as well as theneomycin resistance gene. Maps ofthese plasmids are shownin Fig. 1. Both plasmids were linearized by digestion withrestriction endonuclease Cla I and then transfected intomouse LTK- cells. Cell lines stably resistant to G418 andcontaining one or several copies of pTK2TK1-8 orpTK2TK1-28 integrated into the L cell genome were isolated.The rate of intrachromosomal recombination between thewild-type fragment of the HSV-2 tk gene and the no. 8 or no.28 Xho I linker insertion mutant HSV-1 tk gene was deter-mined by fluctuation tests, using HAT selection to monitortk' segregants. Table 1 shows that we were unable to recoverany recombinants in any of the five cell lines containingpTK2TK1-8 or pTK2TK1-28. The rate of recovery of tk'recombinants between the heterologous tk sequences wasdetermined to be <10-9 events per locus per cell generation.In comparison, the rate of intrachromosomal recombinationbetween a defective internal fragment of the HSV-1 tk geneand the no. 8 mutant HSV-1 tk gene in cell lines containingpAL5 (Fig. 1) is -10-6 events per locus per cell generation(17). [Intrachromosomal recombination between the no. 28mutant HSV-1 tk gene and HSV-1 sequences also occurs ata rate of 10-6 (28).] Constructs pAL5 and pTK2TK1-8 differin two respects; the wild-type tk sequence on pAL5 is 1.2 kbin length and is perfectly homologous to the mutant no. 8 gene(excluding the Xho linker insertion mutation), whereas thewild-type (type 2) tk sequence on pTK2TK1-8 is 800 bp inlength and exhibits 19% base-pair mismatch with the mutantno. 8 gene. From previous work (17), we know that thedifference between 800 bp and 1.2 kb of sequence overlapcannot account for a 1000-fold difference in the rate ofrecombination. Therefore, the .1000-fold change in the rateof recovery of recombinants seen with pTK2TK1-8 com-pared with pAL5 was due to the 19% base-pair mismatchbetween the HSV-1 and HSV-2 tk genes.Extrachromosomal Recombination Between Sequences In-

troduced into Cells by Calcium Phosphate Coprecipitation.

A C B *8 B H H C

neoHSV-I HSV-I

fragment

NOC B Vt B

HSV-Ineo

H H C

HSV-2fragment

c

pAL5

pTK2TKI-8andpTK2TKI-28

kb

B c

FIG. 1. Maps of recombination substrates. BamHI (B), Cla I (C),and HindIII (H) sites are indicated, as well as the positions of XhoI linker insertion mutations 8 and 28. Wavy arrows indicate the 5' to3' orientation of the tk gene sequences. o, ., a, and - representHSV-1, HSV-2, neomycin resistance gene, and vector sequences,respectively. (A) Constructs pAL5 and pTK2TK1-8 (or/andpTK2TK1-28) linearized at the Cla I site. This is the configuration ofthe tk gene sequences when integrated into the L cell genomes in theintrachromosomal recombination studies. (B) Constructs pTK1-8(8.3 kb), pTK1-28 (8.3 kb), pTK2 (6.6 kb), and pAL2 (7.0 kb).

Our inability to recover HSV-1 x HSV-2 recombinants in theintrachromosomal studies presented above was due to a lackof actual recombination between the heterologous genesand/or the formation of nonfunctional hybrid HSV-1/HSV-2tk genes as products of recombination. Because the rate of

Table 1. Effect of 19% base-pair mismatch on the rate ofintrachromosomal recombination

Cells Recombi-Copy tested, nants, Rate of

Cell line no. no. no. recombination8-1 1 2.0x108 0 <1x10-98-2 1 4.6 x 108 0 <1 x 10-98-3 1 4.0x108 0 <1x10-9

28-1 1 4.0 x 108 0 <1 x 10-928-2 3 4.0 x 108 0 <1 x 10-9Cell lines designated 8-1, -2, and -3 contain the construct

pTK2TK1-8 stably integrated into their genomes, whereas cell linesdesignated 28-1 and 28-2 contain the construct pTK2TK1-28 inte-grated into their genomes. Rates of recombination (events pergeneration) between the HSV-1 and HSV-2 tk gene sequences inthese cell lines were estimated from the observed recombinationfrequencies as described. For comparison, the average rate ofrecombination for four lines containing pAL5 (in which the inter-acting tk sequences are perfectly homologous) was previouslydetermined (17) to be 0.8 x 10-6, at least 1000-fold greater than therate ofrecombination between tk sequences exhibiting 19% base-pairmismatch.

Genetics: Waldman and Liskay

5342 Genetics: Waldman and Liskay

extrachromosomal recombination is, in general, much higherthan intrachromosomal recombination (1, 4, 6-9, 32), weexamined extrachromosomal recombination of HSV-1 andHSV-2 tk gene sequences to obtain some evidence thathybrid tk proteins are, indeed, functional.

Plasmid pTK2TK1-8 or pAL5 was introduced into mouseLTK- cells by the calcium phosphate coprecipitate method(26). As shown in Table 2, cells transfected with uncut pAL5(containing homologous tk genes) gave rise to three times asmany HAT colonies as did cells transfected with uncutpTK2TK1-8 (containing heterologous tk genes). The numberof HAT' colonies was proportional to the amount of plasmidDNA transfected, over a 10-fold range of DNA concentra-tion, for both pAL5 and pTK2TK1-8 (data not shown). Thisindicated that the amount of plasmid DNA used was notsaturating and that counting colonies therefore provided avalid determination ofthe relative recombination rates for thehomologous and heterologous tk gene crosses. When pAL5and pTK2TK1-8 were linearized by digestion with Xho Ibefore transfection, the number of recombinants increased-10-fold for both constructs (Table 2). Such stimulation ofextrachromosomal recombination by a double-strand breakin a region of shared homology has been observed in severalprevious studies (4, 5, 12, 33-35).As shown in Table 2, cotransfection of two type 1 tk

sequences on separate plasmids (pAL2 and pTK1-8, Fig. 1)yielded recombinants at a 15-fold greater rate than thatobserved after cotransfection of the heterologous type 1 andtype 2 tk sequences (pTK2 and pTK1-8, Fig. 1). When theplasmid harboring the no. 8 mutant HSV-1 tk gene (pTK1-8)was linearized by digestion with Xho I before cotransfectionwith an homologous (pAL2) or a heterologous (pTK2) part-ner, the rate of recombination was increased 10-fold or20-fold, respectively. The rate of such intermolecular recom-bination exhibited a linear dependence on the amount ofXhoI-cleaved pTK1-8 DNA transfected into the cells, indicatingagain that the number of HATr colonies was not saturatedunder the conditions employed (data not shown).A second type 1 mutant gene, no. 28, was also linearized

by digestion with Xho I and cotransfected into L cells withpAL2 or pTK2. The recombination rate observed with this

Table 2. Effect of 19% base-pair mismatch on the rateof extrachromosomal recombination after calciumphosphate transfection

Dishes, Colonies Ratio,Plasmid(s) total no. per dish (1 x 1):(1 x 2)

Intramolecular recombinationpAL5 18 (3)* 9 3:1pTK2TK1-8 23 (3) 3pAL5/Xho I 5 (2) 117 3-1pTK2TK1-8/Xho I 13 (2) 36

Intermolecular recombinationpAL2 + pTK1-8 9 (3) 8 15-1pTK2 + pTK1-8 23 (3) 0.52pAL2 + pTK1-8/Xho I 8 (3) 112pTK2 + pTK1-8/Xho I 15 (3) 14 8:1pAL2 + pTK1-28/Xho I 8 (3) 126 10:1pTK2 + pTK1-28/Xho I 15 (3) 13

Controlswt HSV-1 tk gene 20 (10) "1000pTK2 5 (1) 0pTK1-8 + pTK1-8/Xho I 5 (1) 0

LTK- cells, 5 X 105 per dish, were transfected with 1 ,g of theindicated construct plus 14 gg of LTK- DNA, or, in the case ofcotransfection experiments, cells were transfected with 10 ,utg of eachconstruct plus no carrier LTK- DNA.*Numbers in parentheses indicate the number of independent ex-periments that were done for each case.

second homologous (1 x 1) cross was 10-fold greater than therate of recombination of the corresponding heterologous (1 x2) cross (Table 2).Extrachromosomal Recombination Between Sequences In-

troduced into Cells by Microinjection. Plasmids pAL5 andpTK2TK1-8 were linearized by digestion with Xho I andmicroinjected directly into the nuclei of LTK- cells, at "20copies per cell. HAT' colonies arose at a 15-fold greater ratefor cells injected with the homologous substrate, pAL5, ascompared with cells injected with the heterologous substratepTK2TK1-8 (Table 3). We estimate that "10o-20% of thecells receiving pAL5/Xho I had undergone a recombinationevent.

Analysis of Recombinant HSV-1 x HSV-2 tk Genes Pro-duced by Extrachromosomal Recombination. To examine thetk genes present in the HATr colonies produced by extra-chromosomal recombination, DNA isolated from severalsuch cell lines was analyzed by Southern blotting (Fig. 2). Outof 17 recombinants examined, 7 were consistent with a geneconversion (or a double exchange) event in which thedefective HSV-2 tk gene sequence donated wild-type infor-mation to the mutant HSV-1 tk gene, thus eliminating the XhoI site and producing a 2.4-kb BamHI fragment resistant toXho I digestion (see Fig. 2).Ten recombinants displayed a 1.9-kb fragment upon diges-

tion with BamHI and HindIII (see Fig. 2), consistent with asingle crossover between the HSV-1 and HSV-2 tk sequencesin which the 5' portion of the recombinant gene is composedof HSV-1 sequences and the 3' portion of the recombinantgene is composed of HSV-2 sequences.

Several [(mutant no. 8) x HSV-2 and (mutant no. 28) xHSV-2] recombinants were examined in greater detail. Forsome (two of four) of the recombinants analyzed that hadapparently undergone a gene conversion, further restrictionanalysis revealed that these tk genes were clearly hybridgenes, each containing a segment of the HSV-2 tk sequencereplacing HSV-1 sequence in a domain encompassing theposition of the insertion mutations. The amount of sequenceinformation that had been transferred was between 300 and800 bp (Fig. 3C). In contrast, the other recombinants thatapparently arose from gene conversion did not contain anyHSV-2 tk restriction sites, indicating that only a small amount(<100 bp) of sequence was transferred in these cases (Fig.3D).

DISCUSSIONAn important question to address regarding homologousrecombination is precisely how well-matched two sequencesmust be in order to undergo recombination. There have beenseveral investigations into the homology requirements ofrecombination in mammalian cells (15-17). These studies,however, all utilized essentially perfectly homologous sub-strates sharing varying lengths of overlapping homology. Wehave begun to address the question of how recombination inmammalian cells is affected by varying degrees of base-pairmismatch. We find that the 19% base-pair mismatch that

Table 3. Effect of 19% base-pair mismatch on the rate ofextrachromosomal recombination of microinjected substrates

Copies Cells HATr Relativeinjected injected, colonies recombination

Construct injected per cell no. scored, no. rate*

Wild-type tk gene 10 446 36pAL5/Xho I 20 2500 38 15pTK2TK1-8/Xho I 20 3020 3 1

*Relative rate is normalized to the frequency of recombinantsfollowing injection of pTK2TK1-8/Xho I, which is ""1 per 1000 cellsinjected.

Proc. Natl. Acad. Sci. USA 84 (1987)

Proc. Natl. Acad. Sci. USA 84 (1987) 5343

1 2 3 4 5 6 7 8 9 10

AATG N So #28 H *8

v Is

K Sm K

ATG N K Sm K

TAG HSV-2FRAGMENT

TGA

SMj1

ATG N So H

ATG N

FIG. 2. Southern blotting analysis of representative tk genes

produced by extrachromosomal recombination. DNA was analyzedusing a probe specific for HSV-1 and HSV-2 tk sequences. Shown isthe analysis of DNA from five recombinants that arose fromextrachromosomal recombination between HSV-2 tk sequences andthe no. 8 mutant HSV-1 tk gene, following calcium phosphate-mediated transfection. Pairwise lanes (e.g., lanes 1 and 2) representindividual recombinants digested with BamHI and HindIII (odd-numbered lanes) or BamHI, HindIII, and Xho I (even-numberedlanes). Indicated are the mobilities of the 2.4-kb fragment diagnosticfor gene conversions and the 1.9-kb fragment diagnostic for singlecrossovers. Also indicated is the mobility of the 800-bp HindIII insert(containing HSV-2 tk sequences) present in the constructs used.Every line shown appears to contain a reconstructed tk gene. Thesamples displayed in lanes 1,2 and 9,10 each arose from a geneconversion; the samples in lanes 5,6 and 7,8 each arose from a singlecrossover; the sample in lanes 3,4 had either undergone both a geneconversion and a crossover event or fortuitously exhibits a fragmentof 2.4 or 1.9 kb. A similar array of recombinants arose fromextrachromosomal recombination between HSV-2 tk sequences andthe no. 28 mutant HSV-1 tk gene (data not shown).

exists between the HSV-1 and the HSV-2 tk genes can reduceintrachromosomal recombination by a factor of at least 1000relative to the rate of intrachromosomal recombination ob-served for two HSV-1 tk gene sequences. This result wasobtained using "crosses" of the same HSV-2 tk sequencewith two different Xho I linker insertion mutants of theHSV-1 tk gene, indicating that this sensitivity to mismatch isnot specific to some particular region of the tk gene.Using the same recombination substrates as were used in

our intrachromosomal studies, we found that extrachromo-somal recombination is considerably less sensitive to base-pair mismatch. When the HSV-1 and HSV-2 tk sequenceswere introduced into L cells on the same DNA molecule oron separate molecules, by calcium phosphate coprecipita-tion, the rate of extrachromosomal recombination was onlyreduced by a factor of 3-15 compared with the extrachro-mosomal recombination rate of comparable homologous tkcrosses. The relative rates of recombination of the homolo-gous versus heterologous tk crosses were not affected bylinearization at the site of the mutation in the HSV-1 tk gene.

Introduction of DNA into mammalian cells by calciumphosphate transfection could result in damage or processingof DNA constructs during passage through the cytoplasm,which might have an effect on recombination. We thereforeperformed experiments in which the constructs were intro-duced by direct microinjection into the nucleus. We foundthat the relative rate of extrachromosomal recombination of

100 bp

FIG. 3. Structures of recombinant tk genes produced by extra-chromosomal recombination. After initial characterization by diges-tion with BamHI, HindIll, and Xho I, as illustrated in Fig. 2, severalrecombinant tk genes were further analyzed by digestion with Hinfl(H), Kpn I (K), Nru I (N), Sac I (Sa), and Sma I (Sm). Shown are

schematic representations of the recombinant gene structures deter-mined. o and - represent HSV-1 and HSV-2 tk gene sequences,respectively. m represents sequence that could not be unambiguous-ly assigned as type 1 or type 2 but that necessarily contains thejunction between type 1 and type 2 sequence in the recombinantgenes. (A) HSV-1 tk gene showing the locations of Xho I linkerinsertion mutations no. 8 and no. 28. (B) HSV-2 tk gene fragmentused in these studies. (C) Recombinant genes arising from apparentgene conversions that corrected the no. 8 or no. 28 mutant. Sequenceinformation of 300-800 bp was transferred from the type 2 sequenceto the mutant type 1 gene in this type of event, replacing a portionof the type 1 gene that had contained the no. 8 or no. 28 mutation.(D) Recombinant gene resulting from correction of the no. 8 HSV-1mutant gene by a short conversion tract (<60 bp, the distance fromthe Hinfl site in HSV-1 tk to the downstream Kpn I site in HSV-2 tk).This gene does not contain HSV-2 restriction sites for any restrictionenzyme tested. Similar recombinant genes in which the no. 28 mutantgene was corrected by a short (<100 bp) conversion tract were

observed. (E) Family of recombinant genes arising from singlecrossovers between the HSV-2 sequence and the no. 8 mutant HSV-1gene. A similar family of genes was produced by single crossoversbetween the HSV-2 sequence and the no. 28 mutant HSV-1 gene.

the heterologous tk cross compared with the homologous tkcross was the same (that is, reduced by a factor of -15) as inthe calcium phosphate coprecipitation experiments. Thisstrongly suggests that this relative insensitivity to base-pairmismatch is a feature of extrachromosomal recombinationper se.

Molecular analysis of several extrachromosomal recombi-nants indicated that a variety of functional hybrid HSV-1/HSV-2 tk genes were produced that conferred HATr whenpresent as a single copy gene in L cells. This substantiatedthat the failure to recover HSV-1 x HSV-2 tk recombinantsintrachromosomally was not solely the result of faulty hybridtk proteins, but rather was the reflection of a reduced rate ofrecombination between the heterologous sequences.The observed difference in sensitivity to base-pair mis-

match displayed by intrachromosomal and extrachromosom-al recombination in mammalian cells has several possibleexplanations. Perhaps recombination ofchromosomal versusextrachromosomal sequences is accomplished by two dis-tinct pathways, requiring different (perhaps overlapping) setsof gene products. In bacteria, it is well documented thatgenetic blocks differentially affect recombination, dependingon the parental configurations of DNA (36, 37). Phage-episomal F-factor recombination is almost completelyblocked by recB- mutations, whereas F-factor-chromosomecrosses are virtually unaffected (36). A second possibility is

B

kbp2.49-I1.9-

TGASml1

HSV-I GENE

C

_ I_

0.8-

D

owE

TGASm

TAGK

i MMM=

momw

Genetics: Waldman and Liskay

5344 Genetics: Waldman and Liskay

that the nature of the substrates exerts an influence on thehomology requirements of mammalian recombination. Theintrachromosomal substrates are presumably coated withhistones and assembled into chromatin, whereas the extra-chromosomal substrates presumably exist as naked DNAmolecules when introduced into the cells. Another differencebetween the substrates is that there could be more availablefree ends of DNA for extrachromosomal recombination thanfor intrachromosomal recombination. It is possible that suchintrinsic characteristics of the recombination substrates, andnot their locations per se (chromosomal versus extrachro-mosomal), determine which recombination pathway will beused. Alternatively, a third possibility is that a single recom-bination pathway (i.e., a single set of gene products) operateson both intra- and extrachromosomal substrates in mamma-lian cells, but the intrinsic characteristics of each substratemay influence the manner in which the substrate interactswith the recombination machinery, and this, in turn, mayaffect the homology requirements.Although it is not presently clear why intrachromosomal

and extrachromosomal recombination have such differenthomology requirements, this difference should be kept inmind when extrapolating results of extrachromosomal re-combination studies to learn about intrachromosomal events,and vice versa. The possibility of mechanistic differencesbetween intra- and extrachromosomal recombination mayalso prove relevant in developing strategies for manipulatingthe genome by targeted recombination because, in a certainsense, targeting involves both intra- and extrachromosomalrecombination.A stringent homology requirement for chromosomal re-

combination as deduced from the studies presented herewould be sufficient to prevent frequent "unwanted" recom-bination between abundant repeated sequences such as theAlu and Kpn I repeats, the family members of which containsequences that exhibit about 80-85% homology (29, 38).Finally, whether the rate of intrachromosomal recombinationis determined primarily by the overall degree of heterology orby the distribution of mismatches can be addressed by usinggene sequences that are more closely matched than thoseused in this study.

The authors thank Dr. Hamish Young, Dr. Kevin Kelley, Dr.Barbara Criscuolo Waldman, Roni Bollag, and Alan Godwin for theirvaluable comments. We also thank Janet Stachelek for her excellenttechnical assistance. This work was supported by National Institutesof Health Grant GM32741 to R.M.L. and P01 CA39238 to Dr. WilliamC. Summers. R.M.L. is a Leukemia Society of America Scholar, andA.S.W. is a Leukemia Society of America Fellow.

1. Rubnitz, J. & Subramani, S. (1986) Mol. Cell. Biol. 6,1608-1614.

2. Folger, K., Thomas, K. & Capecchi, M. R. (1984) Cold SpringHarbor Symp. Quant. Biol. 49, 123-138.

3. Kucherlapati, R. S., Ayares, D., Hanneken, A., Noonan, K.,Rauth, S., Spencer, J. M., Wallace, L. & Moore, P. D. (1984)Cold Spring Harbor Symp. Quant. Biol. 49, 191-197.

4. Lin, F.-L., Sperle, K. & Sternberg, N. (1984) Mol. Cell. Biol.4, 1020-1034.

5. Brenner, D. A., Kato, S., Anderson, R. A., Smigocki, A. C.

& Camerini-Otero, R. D. (1984) Cold Spring Harbor Symp.Quant. Biol. 49, 151-160.

6. de Saint Vincent, B. R. & Wahl, G. M. (1983) Proc. Natl.Acad. Sci. USA 80, 2002-2006.

7. Small, J. & Scangos, G. (1983) Science 219, 174-176.8. Shapira, G., Stachelek, J. L., Letsou, A., Soodak, L. K. &

Liskay, R. M. (1983) Proc. Natl. Acad. Sci. USA 80,4827-4831.

9. Wake, C. T. & Wilson, J. H. (1979) Proc. Natl. Acad. Sci.USA 76, 2876-2880.

10. Volkert, F. C. & Young, C. S. H. (1983) Virology 125,175-193.

11. Liskay, R. M., Stachelek, J. L. & Letsou, A. (1984) ColdSpring Harbor Symp. Quant. Biol. 49, 183-189.

12. Smith, A. J. H. & Berg, P. (1984) Cold Spring Harbor Symp.Quant. Biol. 49, 171-182.

13. Lin, F.-L. & Sternberg, N. (1984) Mol. Cell. Biol. 4, 852-861.14. Stringer, J. R., Kuhn, R. M., Newman, J. L. & Meade, J. C.

(1985) Mol. Cell. Biol. 5, 2613-2622.15. Rubnitz, J. & Subramani, S. (1984) Mol. Cell. Biol. 4,

2253-2258.16. Ayares, D., Chekuri, L., Song, K.-Y. & Kucherlapati, R.

(1986) Proc. Natl. Acad. Sci. USA 83, 5199-5203.17. Liskay, R. M., Letsou, A. & Stachelek, J. L. (1987) Genetics

115, 161-167.18. Singer, B. S., Gold, L., Gauss, P. & Doherty, D. H. (1982)

Cell 31, 25-33.19. Shen, P. & Huang, H. V. (1986) Genetics 112, 441-457.20. Watt, V. M., Ingles, C. J., Urdea, M. S. & Rutter, W. J.

(1985) Proc. Natl. Acad. Sci. USA 82, 4768-4772.21. Mautner, V. & MacKay, N. (1984) Virology 139, 43-52.22. Swain, M. A. & Galloway, D. A. (1983) J. Virol. 46,

1045-1050.23. Kit, S., Kit, M., Qavi, H., Trkula, D. & Otsuka, H. (1983)

Biochim. Biophys. Acta 741, 158-170.24. Graham, F. L. & Van der Eb, A. J. (1973) Virology 52,

456-467.25. Potter, H., Weir, L. & Leder, P. (1984) Proc. Natl. Acad. Sci.

USA 81, 7161-7165.26. Capecchi, M. R. (1980) Cell 22, 479-488.27. Syzbalski, W., Szybalska, E. H. & Ragni, G. (1962) Natl.

Cancer Inst. Monogr. 7, 75-88.28. Liskay, R. M. & Stachelek, J. L. (1986) Proc. Natl. Acad. Sci.

USA 83, 1802-1806.29. Lerman, M. I., Thayer, R. E. & Singer, M. F. (1983) Proc.

Natl. Acad. Sci. USA 80, 3966-3970.30. Southern, P. & Berg, P. (1982) J. Mol. Appl. Genet. 1,

327-341.31. Wagner, M. J., Sharp, J. A. & Summers, W. C. (1981) Proc.

Natl. Acad. Sci. USA 78, 1441-1445.32. Folger, K. R., Thomas, K. & Capecchi, M. R. (1985) Mol.

Cell. Biol. 5, 59-69.33. Brenner, D. A., Smigocki, A. C. & Camerini-Otero, R. D.

(1985) Mol. Cell. Biol. 5, 684-691.34. Kucherlapati, R. S., Eves, E. M., Song, K.-Y., Morse, B. S.

& Smithies, 0. (1984) Proc. Natl. Acad. Sci. USA 81,3153-3157.

35. Anderson, R. A. & Eliason, S. L. (1986) Mol. Cell. Biol. 6,3246-3252.

36. Porter, R. D., McLaughlin, T. & Low, B. (1979) Cold SpringHarbor Symp. Quant. Biol. 43, 1043-1047.

37. Fishel, R. A., James, A. A. & Kolodner, R. (1981) Nature(London) 294, 184-186.

38. Deninger, P. L., Jolly, D. J., Rubin, C. M., Friedmann, T. &Schmid, C. W. (1981) J. Mol. Biol. 151, 17-33.

Proc. Natl. Acad. Sci. USA 84 (1987)